Does anyone have tips and tricks for modelling subs in Fusion (or other CAD)

[Nicoolai]

Hi,
I want to model an old Danish sub, in Fusion 360, and eventually 3D print it.
I've kind of gotten stuck a few times though, not being able to model it.

I've seen some others here, mention that they modelled theirs in Fusion, so hoping for some tips&tricks.

My initial idea was to do sketch planes with the cross-sections and simply loft them.
I've attempted this approach a few times but I always end up with a weird looking geometry.

I've also attempted the sculping functionality but I am way out of my comfort zone there and I can't even get a basic shape that looks anything like what I want.

So, how do you guys do it?

[RanSan]

What sub are you modelling? That might help us give you some tips. Lofting with rails works well once you get the hang of it. Revolve also is a go to for basic shapes.

[CC Clarke]

None of us who became modelers did it by tackling something way beyond our skill set. This is the fastest, most guaranteed way to get frustrated, lose interest and finally give up. Modeling is an art and science as much as an acquired skill. Start with something small, like a common household object or a 20mm, open-sided cube that can be used to calibrate a printer. It's more about combining different tools in succession to deliver the desired result.

Once you master hard surface modeling techniques, then you can move to organic modeling methods which are better suited to most types of submarines.

CC

[Nicoolai]

What sub are you modelling? That might help us give you some tips. Lofting with rails works well once you get the hang of it. Revolve also is a go to for basic shapes.

It's a kobben class, German type 207.

https://en.wikipedia.org/wiki/Kobben-class_submarine

I'm thinking the rails part is perhaps my issue. Atleast I never really got anything great using them.

[jphatton]

I still consider that I have a lot still to learn with CAD modelling, but here are a few suggestions which might help in getting a reasonable approximation to the geometry. This is mainly based on my experience making CAD models with Rhino but the same basic principles should apply to most other 3D CAD software.

- First make sure you have a good handle on the various options for the commands which you are using to generate the surfaces. Most lofting commands have different options for alignment of the curves and the fit (eg. tight, loose etc) between the curves. If you use the wrong combinations you can end up with something very different to what you want...

- For more complex hull forms it is a good idea to break this down into simpler elements rather than trying to generate the entire hull surface in one go. At least in Rhino you can use the edge of one surface either as a rail to guide the subsequent lofted surface or to generate a curve from the edge which can then be used as a rail.

- Probably the easiest way to generate an element which is axisymmetric is by revolving a curve around the axis. The axisymmetric surface can then be cut and split to join to other surfaces if needed.

- I generally lofted hull surfaces using two rails along the edges of the sections, rather than directly between the section curves alone. Usually for submarine the rail line can be derived from the hull edge line in the plan and side view. Also I tend to generate the hull as halves (unless axisymmetric), as these halves can then be easily mirrored across the axis to create the entire hull

- In Rhino there is a network surface command (networksrf) which generates a surface from a set of intersecting curve. I've found this works quite well for generating surfaces such as the curved top of a submarine sail, as well as some front deck areas.

- Don't assume that all of the sections and lines are accurate, even if working from an official drawing. There is often distortion due to scanning and there may be inconsistencies in positioning of items between different views on the original drawing. Therefore, you will need to do some fine adjustment to get hull lines and sections to fit together.

- Check that hull lines and sections curves traced from a drawing are fair - i.e. with smooth curves and transitions. Often there will be analysis commands in the software to measure curvature (eg. curve graph or surface curvature). This allows you to adjust the curves to get smooth curves. If you do not have smooth transitions between sections you will often end up with uneven surfaces. Often you can also simplify curves (ie. reduce numbers of control points) after the initial construction which helps subsequent adjustment.

- curves generated using geometrical functions (eg. circles, arcs, ellipses, parabolas or polynominals) might provide reasonable fits to some hull shapes & by definition should have smooth transitions. NACA 4 series sections often provide good sections for control surfaces - ordinates for these foil sections are available online or can be generated from the equations. Also some CAD software provides plugins which can generate these shapes. For my Thresher/Permit model I got a good approximation to the axisymmetric hull shape using the Myring hull equations (generated using a definition in Grasshopper to create the hull curve and surface).

There are several different ways that a model can be created and other members on the list may have some other methods they can suggest or other import points to consider. In any case it is a good idea to document each step of constructing the model and experiment with different approaches until you end up with something you are happy with. Often I go through several iterations to get to the final model.

[CC Clarke]

All good points in thread #5.

The real questions back to the OP becomes:

1) How much modeling experience do you have with Fusion 360 and 3D modeling in general? Curious and just starting out?

2) Which version of Fusion 360 are you using? The free version is the most popular with hobbyists, but leaves out some capability - like all free (vs trial) versions of apps.

3) If you want to build up expertise, --as in anything else, practice is the key. It can take years to get really, really good. Start small to build your confidence.

4) There is no substitute for butt-in-seat training. For the most part, the days of self-taught "expert" modelers, (those who can find gainful employment with their skills) ended many years ago. If you want to get the fast-track to creating good models, take a class. You can usually get a substantial software discount with a student ID for major software packages. My modeling instructor promised we would learn more in 3 weeks than he learned on his own in three years. He kept that promise too. I had six months of learning on my own and had to un-learn a lot of bad habits. There may be many ways to build something, but there are even more bad ways to build it. Good modeling techniques save more time than they take to learn. The trouble is, in the beginning, it's hard to know what's important to memorize and what isn't.

5) With very few exceptions, there is no need to buy plugins to make up for skills.

6) I've used high-end CAD software, (ProE and SolidWorks) Hollywood-level 3D modeling and animation software and experimented with most of the leading packages in between.

Here are a few tips I've learned over many years:

a) You usually get what you pay for. Freeware, (Blender, Fusion360) have clunky interfaces and limited capability compared to their "paid" cousins. Yeah, if you're on a budget, I get it, and for simple geometric shapes, they're fine, but if you want to build major projects that require a deep understanding of the techniques to be used and the software that can make it happen, then it's worth the time, money, and effort required to learn them.

b) If you are a hobbyist and create minimal content but still want to model something major for your own use, then it's cheaper, (and waaay faster) to pay someone (whether a company or an individual) to build it for you. Even for those of us who create content for a living, if I'm on a time crunch for a project deadline, then I'll shell out a few hundred bucks to get the model (in-hand) in minutes and then spend whatever remaining time I have left, re-modeling or re-texturing texturing it to meet the requirement.

Even some "pro" models are trash too, created by someone who mistakenly believed if it looks good, than it is good. --Nothing could be further from the truth under the harsh light of a digital spotlight. Good models incorporate efficient mesh topology, which yields the best shape with the fewest amount of polygons. Too few polys, and you end up with a lousy model for printing that is covered with faceting that has to be re-worked after printing. Too many polygons and the print can take an eternity to knock out. Many like to blame the printer (or slicer) for crummy-looking models. It's usually the model to blame.

If you've ever tried to have a slicer repair an object prior to slicing and it continuously fails, then it takes time to dig into the mesh and determine where the creator went off the modeling rails. This brings up an important point:

CAD programs can be nightmares for topology since they don't care about it, they rely on the accuracy of the designer's measurements and ability to use the software tools to build to required shapes. Good luck trying to locate embedded polys, and other forms of hosed geometry.

For digital deep-diving, you'll need a 3D modeling program that operates at the point, edge, and polygon levels for super-fine shape adjustment. As much as I enjoy CAD-specific apps, I always return to 3D modeling software to get it as close to perfection as possible, with the same accuracy tolerances during 3D printing verification and evaluation. This is especially true with organic objects - like certain types of submarine hulls. Either way, it can take months of research followed by modeling before a project is completed. I allow six months to ensure adequate R&D + documentation time for each hull.

Depending on how far you are into this, you have been provided with some paths to explore. I've been modeling for over fifteen years professionally, and the learning NEVER ends. I'm still adding new combinations of tools to my workflow.

The bottom line is, the majority of RC submariners don't 3D print. An even smaller minority who own printers know how to model properly for printing, so if this is something you want to aspire to, you're getting in at an ideal time, --depending on your goal.

Invest in the time for training, (YouTube barely qualifies.) Learn and master the various modeling techniques, then graduate to 3D printing which is another subject in itself. Go too fast, and I can almost promise you will fail and find some other way to spend your time. Do it right, and you'll have a blast.

CC

[Nicoolai]

Wow folks, thank you very much for all this detailed information. This is super helpful.

1) How much modeling experience do you have with Fusion 360 and 3D modeling in general? Curious and just starting out?

I do have some experience. I have designed production equipment in the past (though many years ago) and worked a lot with airfoil generation (though mostly through programming the CAD). I imagined it would be no different from making an airfoil, but there are bit more details on the sub :P
Most of my experience though is with good old Inventor and Siemens NX.
This is all 10+ years ago though, now I just make small parts, once in a blue moon, to 3D print (in Fusion).

2) Which version of Fusion 360 are you using? The free version is the most popular with hobbyists, but leaves out some capability - like all free (vs trial) versions of apps.

The free one. This is purely hobby and I am not willing to spend big money on a CAD system that I can't really justify.

With regards to getting someone else to design it, I really want to do that myself. This is part of the process I enjoy and a great learning opportunity.
I've also realized that CAD skills do become rusty, so nice to update those a bit :)

I'm not on a deadline of any kind. If this takes years to complete, that is fine. It is just a hobby for me, something to spend my time on when I happen to have some time available. It is probably the process more than it is the end goal, for me.

[Subculture]

I'm sure there are better pieces of software out there than Fusion and Blender, but none at the price or even a little bit more, and the alternatives tend to come at thousands just for an annual license. No amateur modeller is going to wear that. If it's a living you can justify it as an expense and probably offset the cost against tax and recoup through earnings.

I understand the issues with lofting on more complex shapes, as I'm going through the same learning curve myself. Assuming you're happy with sketching, intersecting and projecting sketches, it's generally once you get into skinning the shape in surfaces things go wonky.

One of the best links I've seen for lofting between profiles and dealing with problems with geometry is this one--

https://forums.autodesk.com/t5/fusion-360-design-validate/lofting-from-intersecting-edges/m-p/6444854#M64601

[DrSchmidt]

I use ViaCad Pro which sells at 700$. That's O.K. for a hobby. I do all my 3D design in it and I'm pretty happy with what it can. Of course it takes experience to learn some tricks, but o far I could achieve what I needed. When it comes to lofts, my problem usually is to get the right cross sections of a boat. Especially complex hulls of WWI and WWII area cann be tricky without having good accurate cross sections.

[Subculture]

Most of the stuff I'm interested in rarely have drawings, at best I'll have a three view to work with. So I need something that can interpolate.

What do you think are the advantages of ViaCad pro over something like Fusion 360 for someone in our hobby? prior to Fusion i did have a go with Rhino 3D, and I found it difficult to get started.

[DrSchmidt]

Never worked with fusion. So I cannot comment on that. I'm self-tought, started with AutoSketch (2D) and then AutoCad. A little bit of Solid Works, but for over 10 years I use ViaCAD.

[jphatton]

Most of the stuff I'm interested in rarely have drawings, at best I'll have a three view to work with. So I need something that can interpolate.

What do you think are the advantages of ViaCad pro over something like Fusion 360 for someone in our hobby? prior to Fusion i did have a go with Rhino 3D, and I found it difficult to get started.

I guess that most CAD software can probably do the job, if they have the key functionalities for working in 3D. Ultimately comes down to the time needed to learn the methodology and tools. I found it took me quite some time to get to the stage of being able to do more complex shapes, but it was also quite satisfying to make progressively more complex models :-)

I initially tried some of the free CAD software but eventually settled on Rhino as at least for me I found it fairly intuitive to learn and had good documentation / tutorials. Like Dr Schmitt, my CAD modelling is purely a hobby & I am self taught.

Often more complex hull shapes take several tries to get right. Here are a few screenshots of an Oberon model I did from a relatively basic plan with some sections, some of which were a bit inaccurate. After drawing sections I had to rework these a bit to get a shape which agreed with the plan and side hull lines. Also there was no detailed bow sections, so eventually this was constructed as a set of networksurfaces (Rhino networksrf command which makes a surface from intersecting curves). The initial version of the bow didn't resemble the real boat closely enough but I could use the preiminary surface to construct a new set of curves which were then adjusted to get something that looked more or less correct.

In terms of errors on this model the transition between a couple of the aft sections isn't as smooth as it should be and I think the bow could be improved a bit further. Also subsequently checking against a more accurate plan I found the sail is a slightly too long and the aft hull starts to curve slightly further forward on the hull than on the model. So the end result isn't perfect but at least looks reasonably close to Oberon.